1
Big Data Healthcare: A Leader's Story
Big Data Symposium: Insights and Actions to Reshape the Healthcare Environment
Session #BG4 | Monday, February 11, 2019
Rasu Shrestha, MD, MBA | EVP, Chief Strategy Officer, Atrium Health
@RasuShrestha
2
Rasu Shrestha, MD, MBA,
EVP, Chief Strategy Officer, Atrium Health
Has no real or apparent conflicts of interest to report.
Conflict of Interest
3
Identify challenges behind successful implementation of big data
analytics for improved healthcare outcomes
Share best practices for big data deployment in a healthcare
setting
Evaluate metrics and outcome measures used to demonstrate
value of big data healthcare implementation
Discuss sustenance models for big data initiatives and future
trajectories
Learning Objectives
Hype Cycle for Emerging Technologies
Hype Hope Here
Data:Base
Technology as an Enabler of Better Care
@RasuShrestha
Zak Kohane, Harvard DBMI
Data blind spots
Existing literature is clear about the
importance of social determinants of health
in improving the health of populations.
These studies uniformly suggest
that nonmedical factors
play a substantially
larger role than do medical
factors in health.
https://bluecrossmafoundation.org/sites/default/files/download/publicatio
n/Social_Equity_Report_Final.pdf
“Leveraging Social Determinants of Health. What Works?” Prepared for the Blue Cross
Blue Shield of Massachusetts Foundation by Yale Global Health Leadership Institute.
Understanding why patients opt out of interventions is a blind spot for many organizations.
Much like how a gardener sows his seeds, and cares for and
nurtures his garden, managing data, especially at scale, requires
some discipline and, arguably, a good deal of passion.
Connecting big data to big insights
@RasuShrestha
17
Source: Analytics: The New Path to Value, a joint MIT Sloan Management Review and IBM Institute of Business Value study. Copyright © Massachusetts Institute of Technology 2010.
Data
Information
Knowledge
Insights
Behavior
Change
Nudges
19
Semantics Harmonizes Data from
Diverse Systems
Data gathered from
diverse sources needs
to be stored and
reviewed in one
consistent form
Data needs to be
“normalized and
standardized
The data’s content
needs to be preserved:
its context in time,
space and in
relationship with other
data allowing EMRs to
remain viable data
sources
UMLS LOINC NDC RxNorm ICD-9-CM
ICD-10-CM ICD-10-PCS DRG APC APDRG
CPT HCFA HCPCS CDT SNOMED CT HL7
HL7 CVX
OMB
Race/Ethnicity
Standards
Commercial
Interface
Terminologies
Provider
Taxonomy
Revenue Codes
Mapping data from different source systems to national
standards
Harmonization between clinical systems
Creation of ontologies to support clinical use cases around
data exchange, meaningful use and disease management
20
Data: building blocks to insights
21
Taming the Big Data Beast
Transformation
Ontology, Logic
Customizability, Design
Harmonization
Semantic Interoperability
Vocabulary
Aggregation
Syntactic Interoperability
Identification
Intelligent
Healthcare
Clinical
Context
Evidence
Based Logic
Role- based
alerts
Analytics at
the point of
care
Person-
Centered
Care
Population
Health
Information Reuse and Integration: Foundation for Accountable Care
Evolution vs Revolution
23
Is Your BI Solution Meeting Your Needs and Driving Tangible
Outcomes?
KLAS Research, Healthcare Business Intelligence 2018
24
KLAS Research, Healthcare Business Intelligence 2018
Broadly Focused vs Niche
Focused
Advanced Analytics
Newcomers Making a
Splash
Open-Source Technology
Threatens Cross-Industry
Behemoths
Highest Performers
Improve Patient Care and
Reduce Costs
25
26
27
Big Data Big Impact
Access to new, diverse data and open datasets are fueling drug
discovery and making clinical trials and research more efficient.
Medical research:
Wearable devices, online diagnostic tools and genetic sequencing
services hold the promise of better informed and engaged patients.
Daily life:
Health systems are investing heavily in technology, including machine
learning, which is proving as effective as or more effective than human
diagnosticians.
The patient
experience:
Telemedicine and health apps make it possible for physicians to see
patients virtually, outside of traditional facilities for increased access
and tailored care.
Ongoing care:
Health data is allowing doctors to build better patient profiles and
predictive models to more effectively anticipate, diagnose and treat
disease.
Prediction and
prevention:
Adapted in part from: Harnessing the Power of Data in Health, Stanford Medicine Health Trend Report
28
Silos and roadblocks prevent effective data sharing but, at the same time,
privacy and security of patient data is paramount.
Data sharing and
security:
Data privacy and interoperability must be addressed at a legislative level to
create a regulatory environment that encourages innovation and research
while putting patients first.
Policy and legislation:
Frustrations with the design of electronic medical records undermine the
physician-patient relationship.
Electronic medical
records (EMRs):
Without proper infrastructure and a data-literate clinical workforce, health
data can only be collected and stored, not leveraged fully.
Skills and training:
Reliance on reactive health care will hamper physicians’ ability to anticipate,
diagnose and treat disease.
Care models:
Big Data: Challenges/ Opportunities
Adapted in part from: Harnessing the Power of Data in Health, Stanford Medicine Health Trend Report
Rhetoric
vs
Reason
@RasuShrestha
@RasuShrestha
42
Big Data: It’s about context
43
BALANCE
Big Data and Big Science Appropriate Variation in Care
Clinical Redesign
DECREASE variation in population care
Personalized Medicine
INCREASE variation in individual care
Big Science “omics”
+
Systems Biology
Go from “Syndrome” to precise individual network
Big Data Analytics
Targeted therapy (reduce unnecessary care)
New Models of Care:
Coordinated, team-based, continuous, accountable, affordable, with aligned incentives
44
Knowledge
Economy
It’s not man vs. machine…
It’s man vs. man and machine
Embrace of digital thus far
has been a replacement of
analogue
With Big Data + AI, we
should go beyond
comparison to humans
Instead, reinvent what it
means to leverage power of
machines at scale, and
augment most humanistic
aspects of care.
@RasuShrestha
What Should We Remember? Too Much Information
Need To Act Fast Not Enough Meaning
Quantitative
AI that produces
complex reports &
documentation. Also
AI in modalities.
Automative
AI that drives
worklists or
diagnostic tasks
Assistive
AI that labels
anatomy; segments;
assists with
diagnosis
Qualitative
AI that qualifies
metrics/ value for
the system across
broader parameters
@RasuShrestha
CheXNet
outperforms the
best published
results on all 14
pathologies in the
ChestX-ray14
dataset.
Core consideration as
Big Data + AI veers
towards ‘escape
velocity’:
Data science training
Focus on delivery
mechanisms and
workflow, not just
apps
Become more
purposeful
Think more
holistically
Use Big Data + AI to
humanize care
@RasuShrestha
@RasuShrestha
T H A N K Y O U
Rasu Shrestha MD MBA
Chief Strategy Officer | Atrium Health
@RasuShrestha